Thanks to everyone who helped me with my transformer selection for my F5T project. I am excited to start building it!
With that problem out of the way, I started to order the whole bunch of PCBs and stuff from the DIYaudio store.
To my disappointment, no matching IRFP240 or IRFP9240 MOSFETs are for sale.
Another road block for now but I am sure my fella DIYers will give me ideas to solve it. 😉
TIA.
With that problem out of the way, I started to order the whole bunch of PCBs and stuff from the DIYaudio store.
To my disappointment, no matching IRFP240 or IRFP9240 MOSFETs are for sale.
Another road block for now but I am sure my fella DIYers will give me ideas to solve it. 😉
TIA.
Last edited:
There is an article for matching Mosfets that can be found on
FIRST WATT and also PassDiy
It's pretty easy.
FIRST WATT and also PassDiy
It's pretty easy.
DiYAudio store has them, matched:
F-5T V1/2 Transistor Kit – diyAudio Store
Only thing is you have to wait for them to become back in stock.
F-5T V1/2 Transistor Kit – diyAudio Store
Only thing is you have to wait for them to become back in stock.
The second ebay listing is by alweit. He has a good reputation and is a member
on this forum. You might try contacting him.
on this forum. You might try contacting him.
I've spent last weekend matching a batch of 60 mosfets (30 x IRFP240, 30 x IRFP9240)that I obtained from Mouser and to my surprise, the tolerances were pretty small. Especially with the IRFP9240 transistors. With those I had only one outlier, and two almost equally sized groups (like 14 and 15) within 10mV. So matching a set of transistors is not that hard.
I was wondering whether there is any value in trying to match the N-channel and P-channel as closely as possible? P-channel seems to have a lower average Vgs for a given drain current than the N-channel mosfets. From the 60 transistors I was only able to pair two outliers (one P-channel, one N-channel) within 20mV (4,82V and 4,81V respectively).
I was wondering whether there is any value in trying to match the N-channel and P-channel as closely as possible? P-channel seems to have a lower average Vgs for a given drain current than the N-channel mosfets. From the 60 transistors I was only able to pair two outliers (one P-channel, one N-channel) within 20mV (4,82V and 4,81V respectively).
> I was wondering whether there is any value in trying to match the N-channel and P-channel as closely as possible?
Yes certainly, if you match everything else, including Vgs, Yfs of MOSFETs, JFETs and source resistors.
Unless of course you like 2nd (and other even order) harmonics.
You can find out more in the F5X thread.
We match everything routinely to < 0.3%, under actual working conditions (voltage, current, temperature, ...).
Patrick
Yes certainly, if you match everything else, including Vgs, Yfs of MOSFETs, JFETs and source resistors.
Unless of course you like 2nd (and other even order) harmonics.
You can find out more in the F5X thread.
We match everything routinely to < 0.3%, under actual working conditions (voltage, current, temperature, ...).
Patrick
Brave one could try place low-noise high-GBW opamp in front of F5.
Doing this would provide not less than 90dB feedback depth through overall audioband with gain of ~35 times and unconditional stability with aperiodic step response.
Attachments
Bes,
you make the composite look too easy.
Any guidance on adapting your topology to other Power Amplifiers?
you make the composite look too easy.
Any guidance on adapting your topology to other Power Amplifiers?
Let's remember mr. Einstein - it must be done as simple as possible, but not simpler.
😉
Any non-inverting current-gain stage can be used inside an opamp's loop for up to ~14V output, then you must use current-gain stage with appropriate voltage gain. The higher voltage gain you could provide the deepeer feedback will be, but you must change correction capacitor values for stability and good aperiodic response.
Input resistance is better to keep not higher than 2 kOhm due to Johnson's noise. Overall scheme is inverting, so mostly immune to common-mode noise.
Input could be easy changed to balanced.
Supply are better to be well common-mode decoupled from AC mains for breaking ground loop through AC mains.
Yes, Papa, you are clearly right and hit the bull's eye!
But taking this exact as preference we could go even further.
Carefully resolving common-case equation around two and four pole compensation we can estimate optimal pole and zero frequencies for audio-preferred aperiodic step response.
Here, for example, possible headamp with a gain of 5, input resistance of 2 kOhm and audio-optimized open-loop gain. All poles moved as high as possible, preferably upper than 20 kHz for providing +-flat depth through audioband.
First three poles provides very fast gain drop, then two zeros compensate gain curve for good stability around unity gain frequency and audio-friendly aperiodic step response, and then the last pole closes the curve.
Shown but not connected input filter have a corner around 80 kHz.
Attachments
It's been quite a while since I last put a compensation cap in an amplifier. I
think it was on the J2, where about 10 pf in the feedback loop cleaned up
the square wave a bit. For a while, I just used a 1 inch length of twisted
Teflon coated wire for this. Cheapest Teflon cap you can get.
think it was on the J2, where about 10 pf in the feedback loop cleaned up
the square wave a bit. For a while, I just used a 1 inch length of twisted
Teflon coated wire for this. Cheapest Teflon cap you can get.
- Status
- Not open for further replies.